INDEX

Explanations

words or numbers after punctuation

The neuron selectively activates on “tokens” that contain non‐lowercase letters—i.e. digits, punctuation, or uppercase characters—typical of code symbols or numeric identifiers.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

sprzedam

-1.96

 ventes

-1.93

 Mereka

-1.82

쐈

-1.80

 autonomie

-1.80

 materiale

-1.77

 الأولى

-1.76

삵

-1.75

 horloge

-1.74

 ถูก

-1.74

POSITIVE LOGITS

</strong>

3.31

</em>

2.42

2.38

2.34

2.23

2.14

1.99

1.95

ING

1.95

 However

1.92

Activations Density 0.000%